Archive · 12 Sep 2026
Anthropic CEO Urges Slower AI Development Amid Superintelligence Risks
Dario Amodei, the head of Anthropic, wants AI companies to slow down a little.
He says AI will still improve quickly even if developers take more care.
The extra time could help people make better safety rules.
Amodei says powerful AI could bring important benefits.
He also warns that developing it too quickly could be dangerous.
Jacob Coxon, a researcher who worked at OpenAI and Anthropic, left the industry after making similar warnings.
He said companies were racing toward machines that might improve themselves.
Other AI leaders and workers have also called for a more deliberate pace.
Anthropic CEO Dario Amodei urged AI companies to slow the pace of model capability improvements.
Amodei said progress would still feel fast and that the extra time should be used to address serious risks.
Former OpenAI and Anthropic researcher Jacob Coxon accused both companies of recklessly pursuing self-improving superintelligence.
Amodei said refusing to build AI could deny benefits or leave the technology to authoritarian powers, while moving too quickly was reckless.
The comments follow similar calls from Anthropic, OpenAI CEO Sam Altman and more than 1,000 AI-company employees to pace frontier AI development.
- Who
- Anthropic CEO Dario Amodei, former researcher Jacob Coxon, OpenAI CEO Sam Altman and more than 1,000 AI-company employees are central to the discussion.
- What
- Amodei called for slowing the development of increasingly capable AI models because of risks associated with superintelligence.
- When
- Amodei made the statement on Saturday, September 12, 2026, after earlier calls in June and July to slow AI development.
- Where
- The statements concerned AI companies and frontier AI development, primarily in the United States.
- Why
- Amodei said AI risks are serious and that the time gained by slowing down should be used to address them, while acknowledging that not building AI also carries costs.
Arguments for Slowing Development
Arguments for Continuing Development
Safety versus benefits
Arguments for Slowing Development
Developers should slow progress to give society time to strengthen safeguards and address serious risks from powerful AI.
Arguments for Continuing Development
Not building the technology could deprive humanity of its benefits or leave AI development to authoritarian powers.
Pace of the AI race
Arguments for Slowing Development
Coxon argues that OpenAI and Anthropic are racing toward self-improving superintelligence irresponsibly and could endanger humanity.
Arguments for Continuing Development
Amodei argues for a middle way: continue developing AI, but avoid moving so quickly that development becomes reckless.
- Main speaker
- Dario Amodei, co-founder and CEO of Anthropic
- Central proposal
- Slow the pace at which AI models improve their capabilities
- Key risk
- Loss of control over powerful or self-improving AI systems
- Researcher warning
- Jacob Coxon accused OpenAI and Anthropic of “gambling with our lives”
- Earlier industry call
- More than 1,000 employees at cutting-edge AI companies signed a petition supporting deliberate pacing
- Related statement
- OpenAI CEO Sam Altman said developers might need to voluntarily slow advances so society could catch up
- Definition
- Superintelligence is described as a theoretical point when AI capabilities exceed human intelligence